From TF-IDF to Transformers: A Comparative and Ensemble Approach to Sentiment Classification

In three linesComparative study of sentiment classification models on IMDb: Naive Bayes, Logistic Regression, SVM, LightGBM, LSTM, RoBERTa, and DistilBERT. RoBERTa achieves 93.02% accuracy. Soft voting ensemble improves performance.

Summary generated by Claude — human-verified